The Authors 2012. This article is published with open access at S *Corresponding author email Article Atmospheric Science doi 10.1007/s11434-012-5015-4 Satellite remote sensing of changes in NOxemissions over China during 1996–2010 ZHANG Qiang1*, GENG GuanNan1,2, WANG SiWen2, RICHTER Andreas32 State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China; 3 Institute of Environmental Physics, University of Bremen, Bremen 28359, Germany Received October 31, 2011; accepted December 27, 2011 Satellite derived NO2column data have been used to study Chinese national fossil fuel consumption and pollutant emissions. Based on NO2retris from two satellites GOME and SCIAMACHY for 1996–2010, we analyzed the characteristics and evolution of regional pollution related to NOxemissions in China. Satellite observations indicated that the highly polluted regions were expanding. Anthropogenic emission dominated areas have expanded from the east to central and western China, and new highly polluted regions have ed throughout the nation. Bottom-up emission estimates suggested a 133 increase in anthro-pogenic NOxemissions in East Central China during 1996 to 2010, which was lower than the 184 increase of the NO2columns measured by the satellites. We found that growth rates of NOxemissions have slowed in Chinese megacities over recent years, in contrast to which, the NOxemissions were soaring in medium-sized cities, indicating that strict controls of NOxemissions from coal-fired facilities are required in China. NOx, satellite remote sensing, SCIAMACHY, regional pollution, urban expansion Citation Zhang Q, Geng G N, Wang S W, et al. Satellite remote sensing of changes in NOxemissions over China during 1996–2010. Chin Sci Bull, doi 10.1007/ s11434-012-5015-4 Nitrogen oxides NOx NONO2 are major contributors to regional air pollution. NOxis released into the atmosphere as a result of anthropogenic e.g. fossil fuel combustion and natural e.g. soil emissions, lightning, biomass burning processes. They are important precursors of tropospheric ozone and aerosols, and also participate in the ation of acidic precipitation; hence, they are detrimental to human health and the ecosystem. During the past two decades, China has experienced rapid economic growth, converting the world’s largest agrarian nation into an industrial society. This process is driven by fossil fuel consumption, and large volumes of air pollutants are released into the atmosphere, leading to a series of complex air pollution problems such as acid rain, haze, and photochemical smog [1,2]. NOxemission has a close relationship with fossil fuel consump-tion, and subsequently in China over the last 20 years, NOxhas increased the most rapidly of any air pollutant [3–5]. An accurate understanding of the complex regional air pollution largely depends on the quantification of a priori emissions. Previous studies focused on emission budgets and spatial-temporal distributions of air pollutants, are mainly based on bottom-up emission inventory data [3,5,6]. How-ever, this is only as accurate as the statistical data and local emission factors [5] used in the calculation. Re-cently developed space-borne remote sensing technology provides an alternative and effective approach to the quanti-fication of air pollutants [7,8]. Among the various species amenable to satellite observation, NO2has been studied the most broadly [9–13]. This is because 1 retri of NO2is less affected by the other strongly absorbing atmospheric species; and 2 NO2has a short atmospheric lifetime, which allows closer linking of the measured NO2columns to the 2 Zhang Q, et al. Chin Sci Bull January 2012 Vol.57 No. surface emissions of NOx[10,14,15]. The anthropogenic NOxemissions are mainly produced alongside energy use, so that NO2in polluted regions is a good indicator of fossil fuel consumption [16]. In this study, we used satellite derived tropospheric NO2columns to study the spatial-temporal variations of NOxemissions between 1996 and 2010 in China, and investigat-ed the forcing factors of these changes. 1 Data and s 1.1 Instruments and data Since 1996, several space-borne instruments have been launched to observe tropospheric NO2columns, including the Global Ozone Monitoring Experiment GOME, 1996– 2002, the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography SCIAMACHY, 2003 to present, the Ozone Monitoring Instrument OMI, 2005 to present, and the Global Ozone Monitoring Experiment-2 GOME-2, 2007 to present. GOME, SCIAMACHY, and GOME-2 have local passing times over China between 930 and 1030 am, and, partly because of the similar algorithms used to retrieve the data, these measurements have been shown to have high consistency [10]. However, serious instrument degradation in GOME-2 since 2009 means it cannot now be used for trend analysis [17]. Although OMI has the higher spatial resolution of the four instruments, it has an early afternoon local passing time, which makes it difficult to directly com-pare to the others. In this work GOME measurements from April 1996 to the end of 2002 were used, and SCIAMACHY measurements from 2003 to 2010. GOME and SCIAMACHY are both sponsored by the European Space Agency. GOME was launched in April 1995 on board the ERS-2 satellite and was the first space- borne instrument designed to measure the precursors of lower tropospheric ozone and other trace gases. SCIAMACHY was launched in March 2002 on board the ENVISAT-1 plat. Both of these instruments are passive remote sens-ing spectrometers. GOME measures the earthshine radiance and the solar irradiance in the UV/VIS spectral range 240– 790 nm at a spectral resolution of 0.2–0.4 nm and a nadir pixel size of 320 km40 km. With a 960 km across-track swath width, the global coverage by GOME can be achieved per three days daily coverage achieved above 65 latitudes [18]. SCIAMACHY measures the spectral ranged from 240 to 2380 nm, covering the near-infrared wavelength. The nadir pixel size for SCIAMACHY is about 60 km30 km, and the spectral resolution is 0.22–1.48 nm. Global cover-age is achieved every 6 days by SCIAMACHY [19]. 1.2 ology In this study, we use the GOME and SCIAMACHY tropo-spheric NO2column products retrieved by the Institute of Environmental Physics, Bremen University. The retri process consists of three steps [10] First, a slant NO2col-umn density is determined from a spectral fit using a Dif-ferential Optical Absorption Spectroscopic DOAS ap-proach at 425–450 nm. Then the stratospheric contribution to the slant columns is estimated by assimilating slant col-umns provided by the SLIMCAT model and then removed from the total slant column. Finally, the residential tropo-spheric slant column is converted into vertical column by application of the tropospheric air mass factor AMF cal-culated by SCIATRAN. For individual retris, the abso-lute error attributed to the random spectral fit and subtrac-tion of the stratospheric contribution is estimated to be 0.5–1.01015molecules cm2. The relative error in the month-average column is 40–60 in polluted regions, largely because of the AMF uncertainty not important for trend analyses. The overall error of the annual change in the NO2column over China is estimated to be 15 [10]. The accuracy of the satellite measurements is reduced by clouds, which block the observation of NO2beneath them. The satellite pixels showing cloud fraction 0.2 were fil-tered out in our analysis. Cloud ination is synchronously derived from the observations of GOME and SCIAMACHY, which is available from the website of the Royal Nether-lands Meteorological Institute http//www.temis.nl/fresco/. We then allocated the swath data into 0.1250.125 hori-zontal grids and calculated the month-average NO2columns for each grid. NOxhas a shorter lifetime in summer so that will be transported less far from sources than in winter. Therefore, the observed NO2columns have the closest rela-tionship to surface NOxemissions in summer [4]. For this reason we used the summer average NO2columns to demon-strate the spatial characteristics of the emissions, and to an-alyze the temporal trend based on the annual average NO2columns. 2 Results 2.1 Spatial-temporal variations of NO2columns during 1996–2010 Figure 1 shows the spatial-temporal variations of NO2col-umns observed by the satellites during 1996–2010 in East Central China. Using only summer satellite measurements, these observations show significant characteristics of regional pollution, with greater spatial coverage of highly polluted areas. During 1996–1998, these regions of high NOxemis-sions were mainly concentrated in the North China Plain, Yangtze River Delta, and Pearl River Delta. However, by 2008–2010, the NOxemission strengths in these areas had increased, resulting in super-regions of pollution including Beijing-Tianjin-Tangshan, Central of Hebei, West of Shan-dong, and North Central of Henan. The extension of high polluted regions in Yangtze River Delta also enlarged. In addition, new hotspots appeared in Jilin, Central Liaoning, Zhang Q, et al. Chin Sci Bull January 2012 Vol.57 No. 3 Figure 1 Satellite observed spatial-temporal variations of NO2columns during 1996 to 2010. Solid lines within the Chinese national boundaries denote the twelve key regions defined in the “Joint Prevention and Control Strategy” by Ministry of Environmental Protection MEP. a Summer average tropospheric NO2columns retrieved from GOME for the period 1996 to 1998. b Summer average tropospheric NO2columns retrieved from SCIAMACHY for the peri-od 2003 to 2005. c Summer average tropospheric NO2columns retrieved from SCIAMACHY for the period 2008–2010. d Ratios of c and b. Grids with tropospheric NO2columns less than 1015molecules cm2are not colored in d. Inner Mongolia, North Central Shanxi, Guanzhong Zone in Shaanxi, the Wuhan City cluster, Chengdu-Chongqing, and the Urumqi City cluster. Figure 2 shows the changes in tropospheric NO2columns between 2000 and 2010 in three differently sized city-clus- ters Beijing-Tianjin-Tangshan, Wuhan and surroundings, and the city-cluster in South Shandong. By 2010, the high NO2columns covered the entire urban areas in these three regions. We also found that the growth rates of NOxemis-sions and pollution levels in medium-sized cities such as Zaozhuang, Linyi, and Jining were comparable to those in the megacities, indicating the urgency of implementing controls on NOxemissions in medium-sized Chinese cities. With the increase of serious regional pollution in China, the General Office of State Council published the “Guiding Opinions on Facilitating the Joint Prevention and Control of Air Pollution and Improving the Regional Air Quality”. This was issued by nine ministries including MEP, and aims to facilitate the joint control of regional pollution in nine later 12 key regions [20]. Figure 1 shows the three eco-nomic circles Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta, and the nine city-clusters Central of Liaoning Province, Shandong Peninsula, the Wuhan City cluster, Changsha-Zhuzhou-Xiangtan, Chengdu-Chongqing, West Coast of Taiwan Strait, West and Central Shanxi Province, Guanzhong zone in Shaanxi Province, and the Urumqi City cluster. Satellite observations between 2008 and 2010 suggested that these key regions marked by MEP in the “Joint Control and Prevention Strategy” cover the main high NOxemission areas. Therefore, these key regions are representative; however, the highly polluted city-clusters in Central of Henan and West of Shandong are not included in the Strategy, and the rapid growth in NOxemissions from industrial regions in Jilin and Inner Mongolia risks turning these two provinces into new highly polluted areas. We suggest considering these regions as top priorities in the 4 Zhang Q, et al. Chin Sci Bull January 2012 Vol.57 No. Figure 2 Changes in tropospheric NO2columns and the major socio-economic inds of three city clusters between 2000 and 2010. From top to bottom Beijing-Tianjin-Tangshan, Wuhan and surroundings, and the city-cluster in South Shandong. next stage of the strategy. In the past decade, the growth rates of NO2columns in Beijing-Tianjin-Hebei, Shandong Peninsula and Yangtze River Delta stand out among these key regions Figure 3. Correlation analysis between the NO2columns and the pop-ulation density indicates a slowing down of the growth rates of the NO2columns in the megacities during 2008–2010. This contrasts with the rapid growth of the same in medium sized cities. Figure 4 shows the profiles and evolution of the summer average tropospheric NO2columns over three typ-ical latitude bands. The growth rate of NO2columns be-tween 2003 and 2010 is smaller than that between 1997 and 2003 over Shanghai, but the high polluted areas have ex-tended. In contrast, the growth rates of NO2columns in me-dium sized cities such as Wuhan, Handan, Zibo, and Zunyi between 2003 and 2010 are larger than those between 1997 and 2003. Similar results are also found over other latitude bands, which can be characterized as the growth of NO2columns in megacities mainly occurred in the first half of the period during 1996–2010, while the NO2columns in medium sized cities have increased more significantly in recent years. It is worth noting that direct comparison of GOME and SCIAMACHY at high-resolution grid level are qualitative because of the differences in spatial resolution between two instruments. Figure 1d shows the growth rates of NOxemissions in different regions over the past five years. To minimize un-certainties, the grids with summer average NO2columns lower than 1015molecules cm2have been removed. We found the most rapid growth of NOxemissions occurred in the old industrial regions in Jilin and Liaoning, and newly developing industrial regions in Inner Mongolia, Shanxi, and Ningxia. The growth of NOxemissions has slowed down in megacities such as Beijing, Shanghai, and Guang-zhou. In regions dominated by anthropogenic emissions, trop-ospheric NO2columns show winter maxima, because of the longer lifetime of NOxand the less turbulent conditions in winter. In regions dominated by emissions from natural sources such as soils and lightning, the emission strength is Zhang Q, et al. Chin Sci Bull January 2012 Vol.57 No. 5 Figure 3 Correlations between the NO2columns and the population density in twelve key regions during 1997-2010. Data are at 0.5 0.5 grids. Colors denote different regions. Grey data are for grids outside the twelve key regions. Figure 4 Profiles and evolution of summer average tropospheric NO2columns in three typical latitude bands during 1997–2010. far higher in